Development of voice recognition for student attendance system is beneficial in many ways. It helps the lecturer in administrative the attendance of their student with efficiency. This is because students always cheat with their attendancy by signing on behalf of their friend who did not attend class. With this project, voice biometric is used as a medium for student to mark their attendance. Cheating among students will be prevented because like fingerprints, each voice is different. The objectives of this project are to study and understand the properties understand the properties of speaker recognition and to analyze the effectiveness of using Euclidean distance feature for speaker recognition. Databases of 26 volunteers were collected consisting of only male. The report result is tabulated. Three types of analysis were done, first same train is used as test data reported 100% correct. The remaining two analyses used different test data recording. Volunteers use the same sentence as test data reported 76.92% correct. Lastly volunteers used their name and the correct percentage is 46.15%.